N-acetylaspartic acid (NAA), a prominent peak in the proton spectrum, is an amino acid thought to be present almost exclusively in neurons and their dendritic and axonal extensions. While 1H MRS studies are showing promise in identifying NAA deficits in different patient groups, unwanted lipid signal from subcutaneous fat surrounding the skull, and necessarily large voxels have limited investigators' ability to assess NAA in cortical gray matter. Here we report a technique developed to derive estimates of NAA signal from cortical gray matter. This approach uses an inversion recovery imaging pulse sequence with a long TE to suppress lipid signal from the scalp and information from concurrently obtained structural MR images to determine the CSF, white and gray matter composition of each spectroscopic voxel. A regression analysis is then used to estimate what NAA levels would be in "pure" white and gray matter voxels. This technique has been applied to demonstrate reduced NAA gray/white levels in the brains of five healthy older compared with five healthy younger women.